To find the maximum or minimum of a function f(x, y) subject to a constraint g(x, y) = c, we search for points where their gradient vectors are parallel. This condition is expressed by the equation: